Trace-penalty minimization for large-scale eigenspace computation
The eigenspace corresponding to the \(k\) smallest eigenvalues of a large-scale matrix pencil \((A,B)\) with symmetric matrices of size \(n\times n\) (and \(B\) positive definite) is the solution of the constrained problem \(\min_X \mathrm{tr}(X^TAX)\) with \(X\in\mathbb{R}^{n\times k}\) satisfying \(X^TBX=I\). It is proposed to solve this problem by minimizing \(f_\mu(X)=\frac{1}{2}\mathrm{tr}(X^TAX)+\frac{\mu}{4}\|X^TBX-I\|_F\) using standard methods. A key result shown is that \(\mu\) need not be very large, but that if suffices to choose \(\max(0,\lambda_k)<\mu<\lambda_n\) where \(\lambda_1\leq\lambda_2\leq\cdots\leq\lambda_n\) are the sorted eigenvalues of \(A\) for there to be essentially only one minimum that is global. Algorithmic aspects of a classical steepest descent minimization procedure are discussed. This includes restarts, the choice of the parameter \(\mu\), and stopping criteria. Also the complexity of the algorithm is discussed and it is extensively tested on numerical examples.
- Trace minimization method via penalty for linear response eigenvalue problems
- The trace minimization method for the symmetric generalized eigenvalue problem
- On generalizing trace minimization principles
- A truncated-CG style method for symmetric generalized eigenvalue problems
- Computational Science – ICCS 2005
- A block Chebyshev-Davidson method with inner-outer restart for large eigenvalue problems
- A Chebyshev–Davidson Algorithm for Large Symmetric Eigenproblems
- A Nonmonotone Line Search Technique and Its Application to Unconstrained Optimization
- A Nonmonotone Line Search Technique for Newton’s Method
- A Trace Minimization Algorithm for the Generalized Eigenvalue Problem
- ARPACK Users' Guide
- Block Krylov-Schur method for large symmetric eigenvalue problems
- Block Locally Optimal Preconditioned Eigenvalue Xolvers (BLOPEX) in Hypre and PETSc
- scientific article; zbMATH DE number 5060482 (Why is no real title available?)
- KSSOLV -- a MATLAB toolbox for solving the Kohn-Sham equations
- Nearly Optimal Preconditioned Methods for Hermitian Eigenproblems Under Limited Memory. Part II: Seeking Many Eigenvalues
- Numerical methods for electronic structure calculations of materials
- On the nonmonotone line search
- Optimization theory and methods. Nonlinear programming
- PRIMME: preconditioned iterative multimethod eigensolver -- methods and software description
- ScaLAPACK Users' Guide
- Toward the optimal preconditioned eigensolver: Locally optimal block preconditioned conjugate gradient method
- Two-Point Step Size Gradient Methods
- Variational methods for the solution of problems of equilibrium and vibrations
- Orbital minimization method with \(\ell^{1}\) regularization
- A Riemannian conjugate gradient method for optimization on the Stiefel manifold
- Spectral residual method for nonlinear equations on Riemannian manifolds
- Trace minimization method via penalty for linear response eigenvalue problems
- Triangularized orthogonalization-free method for solving extreme eigenvalue problems
- Transportless conjugate gradient for optimization on Stiefel manifold
- The space decomposition method for the sum of nonlinear convex maximum eigenvalues and its applications
- A brief introduction to manifold optimization
- Improved SVRG for finite sum structure optimization with application to binary classification
- Nonconvex and nonsmooth optimization with generalized orthogonality constraints: an approximate augmented Lagrangian method
- Implicit steepest descent algorithm for optimization with orthogonality constraints
- KSSOLV 2.0: an efficient Matlab toolbox for solving the Kohn-Sham equations with plane-wave basis set
- scientific article; zbMATH DE number 4016066 (Why is no real title available?)
- An Augmented Lagrangian Method for $\ell_{1}$-Regularized Optimization Problems with Orthogonality Constraints
- A stochastic variance reduction method for PCA by an exact penalty approach
- A conjugate gradient method for electronic structure calculations
- A New First-Order Algorithmic Framework for Optimization Problems with Orthogonality Constraints
- Parallel implementations of the trace minimization scheme trace{min} for the sparse symmetric eigenvalue problem
- A class of smooth exact penalty function methods for optimization problems with orthogonality constraints
- Parallelizable Algorithms for Optimization Problems with Orthogonality Constraints
- An efficient Gauss-Newton algorithm for symmetric low-rank product matrix approximations
- Stochastic Gauss-Newton algorithms for online PCA
- Slow and finite-time relaxations to m-bipartite consensus on the Stiefel manifold
- Weighted Trace-Penalty Minimization for Full Configuration Interaction
- A randomized feasible algorithm for optimization with orthogonal constraints
- Solving weighted orthogonal Procrustes problems via a projected gradient method
- A generalized complex power iteration-based algorithm for quartic-quadratic optimization problems over a spherical constraint and its applications
- Direct minimization on the complex Stiefel manifold in Kohn-Sham density functional theory for finite and extended systems
- A non-monotone adaptive scaled gradient projection method for orthogonality constrained problems
This page was built for publication: Trace-penalty minimization for large-scale eigenspace computation
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2398478)
